X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/722fbc81b200416304b08c9c11a8447131ef765d..2693a197ea37490a891c54bbb279937a7b4e897c:/sdk/python/tests/test_keep_client.py diff --git a/sdk/python/tests/test_keep_client.py b/sdk/python/tests/test_keep_client.py index 5cba8cccbe..7a227f122f 100644 --- a/sdk/python/tests/test_keep_client.py +++ b/sdk/python/tests/test_keep_client.py @@ -1,37 +1,64 @@ +# Copyright (C) The Arvados Authors. All rights reserved. +# +# SPDX-License-Identifier: Apache-2.0 + +from __future__ import absolute_import +from __future__ import division +from future import standard_library +standard_library.install_aliases() +from builtins import str +from builtins import range +from builtins import object import hashlib import mock import os import pycurl import random import re +import shutil import socket -import threading +import sys import time import unittest -import urlparse +import urllib.parse + +import parameterized import arvados import arvados.retry -import arvados_testutil as tutil -import keepstub -import run_test_server +import arvados.util +from . import arvados_testutil as tutil +from . import keepstub +from . import run_test_server -class KeepTestCase(run_test_server.TestCaseWithServers): +from .arvados_testutil import DiskCacheBase + +@parameterized.parameterized_class([{"disk_cache": True}, {"disk_cache": False}]) +class KeepTestCase(run_test_server.TestCaseWithServers, DiskCacheBase): + disk_cache = False MAIN_SERVER = {} KEEP_SERVER = {} + block_cache_test = None @classmethod def setUpClass(cls): super(KeepTestCase, cls).setUpClass() run_test_server.authorize_with("admin") cls.api_client = arvados.api('v1') + cls.block_cache_test = DiskCacheBase() cls.keep_client = arvados.KeepClient(api_client=cls.api_client, - proxy='', local_store='') + proxy='', local_store='', + block_cache=cls.block_cache_test.make_block_cache(cls.disk_cache)) + + @classmethod + def tearDownClass(cls): + super(KeepTestCase, cls).setUpClass() + cls.block_cache_test.tearDown() def test_KeepBasicRWTest(self): self.assertEqual(0, self.keep_client.upload_counter.get()) foo_locator = self.keep_client.put('foo') - self.assertRegexpMatches( + self.assertRegex( foo_locator, '^acbd18db4cc2f85cedef654fccc4a4d8\+3', 'wrong md5 hash from Keep.put("foo"): ' + foo_locator) @@ -40,15 +67,15 @@ class KeepTestCase(run_test_server.TestCaseWithServers): self.assertEqual(6, self.keep_client.upload_counter.get()) self.assertEqual(0, self.keep_client.download_counter.get()) - self.assertEqual(self.keep_client.get(foo_locator), - 'foo', + self.assertTrue(tutil.binary_compare(self.keep_client.get(foo_locator), + b'foo'), 'wrong content from Keep.get(md5("foo"))') self.assertEqual(3, self.keep_client.download_counter.get()) def test_KeepBinaryRWTest(self): - blob_str = '\xff\xfe\xf7\x00\x01\x02' + blob_str = b'\xff\xfe\xf7\x00\x01\x02' blob_locator = self.keep_client.put(blob_str) - self.assertRegexpMatches( + self.assertRegex( blob_locator, '^7fc7c53b45e53926ba52821140fef396\+6', ('wrong locator from Keep.put():' + blob_locator)) @@ -57,33 +84,33 @@ class KeepTestCase(run_test_server.TestCaseWithServers): 'wrong content from Keep.get(md5())') def test_KeepLongBinaryRWTest(self): - blob_str = '\xff\xfe\xfd\xfc\x00\x01\x02\x03' - for i in range(0,23): - blob_str = blob_str + blob_str - blob_locator = self.keep_client.put(blob_str) - self.assertRegexpMatches( + blob_data = b'\xff\xfe\xfd\xfc\x00\x01\x02\x03' + for i in range(0, 23): + blob_data = blob_data + blob_data + blob_locator = self.keep_client.put(blob_data) + self.assertRegex( blob_locator, '^84d90fc0d8175dd5dcfab04b999bc956\+67108864', ('wrong locator from Keep.put(): ' + blob_locator)) self.assertEqual(self.keep_client.get(blob_locator), - blob_str, + blob_data, 'wrong content from Keep.get(md5())') @unittest.skip("unreliable test - please fix and close #8752") def test_KeepSingleCopyRWTest(self): - blob_str = '\xff\xfe\xfd\xfc\x00\x01\x02\x03' - blob_locator = self.keep_client.put(blob_str, copies=1) - self.assertRegexpMatches( + blob_data = b'\xff\xfe\xfd\xfc\x00\x01\x02\x03' + blob_locator = self.keep_client.put(blob_data, copies=1) + self.assertRegex( blob_locator, '^c902006bc98a3eb4a3663b65ab4a6fab\+8', ('wrong locator from Keep.put(): ' + blob_locator)) self.assertEqual(self.keep_client.get(blob_locator), - blob_str, + blob_data, 'wrong content from Keep.get(md5())') def test_KeepEmptyCollectionTest(self): blob_locator = self.keep_client.put('', copies=1) - self.assertRegexpMatches( + self.assertRegex( blob_locator, '^d41d8cd98f00b204e9800998ecf8427e\+0', ('wrong locator from Keep.put(""): ' + blob_locator)) @@ -91,40 +118,56 @@ class KeepTestCase(run_test_server.TestCaseWithServers): def test_unicode_must_be_ascii(self): # If unicode type, must only consist of valid ASCII foo_locator = self.keep_client.put(u'foo') - self.assertRegexpMatches( + self.assertRegex( foo_locator, '^acbd18db4cc2f85cedef654fccc4a4d8\+3', 'wrong md5 hash from Keep.put("foo"): ' + foo_locator) - with self.assertRaises(UnicodeEncodeError): - # Error if it is not ASCII - self.keep_client.put(u'\xe2') + if sys.version_info < (3, 0): + with self.assertRaises(UnicodeEncodeError): + # Error if it is not ASCII + self.keep_client.put(u'\xe2') - with self.assertRaises(arvados.errors.ArgumentError): - # Must be a string type + with self.assertRaises(AttributeError): + # Must be bytes or have an encode() method self.keep_client.put({}) -class KeepPermissionTestCase(run_test_server.TestCaseWithServers): + def test_KeepHeadTest(self): + locator = self.keep_client.put('test_head') + self.assertRegex( + locator, + '^b9a772c7049325feb7130fff1f8333e9\+9', + 'wrong md5 hash from Keep.put for "test_head": ' + locator) + self.assertEqual(True, self.keep_client.head(locator)) + self.assertEqual(self.keep_client.get(locator), + b'test_head', + 'wrong content from Keep.get for "test_head"') + +@parameterized.parameterized_class([{"disk_cache": True}, {"disk_cache": False}]) +class KeepPermissionTestCase(run_test_server.TestCaseWithServers, DiskCacheBase): + disk_cache = False MAIN_SERVER = {} - KEEP_SERVER = {'blob_signing_key': 'abcdefghijk0123456789', - 'enforce_permissions': True} + KEEP_SERVER = {'blob_signing': True} + + def tearDown(self): + DiskCacheBase.tearDown(self) def test_KeepBasicRWTest(self): run_test_server.authorize_with('active') - keep_client = arvados.KeepClient() + keep_client = arvados.KeepClient(block_cache=self.make_block_cache(self.disk_cache)) foo_locator = keep_client.put('foo') - self.assertRegexpMatches( + self.assertRegex( foo_locator, r'^acbd18db4cc2f85cedef654fccc4a4d8\+3\+A[a-f0-9]+@[a-f0-9]+$', 'invalid locator from Keep.put("foo"): ' + foo_locator) self.assertEqual(keep_client.get(foo_locator), - 'foo', + b'foo', 'wrong content from Keep.get(md5("foo"))') # GET with an unsigned locator => NotFound bar_locator = keep_client.put('bar') unsigned_bar_locator = "37b51d194a7513e45b56f6524f2d51f2+3" - self.assertRegexpMatches( + self.assertRegex( bar_locator, r'^37b51d194a7513e45b56f6524f2d51f2\+3\+A[a-f0-9]+@[a-f0-9]+$', 'invalid locator from Keep.put("bar"): ' + bar_locator) @@ -148,72 +191,9 @@ class KeepPermissionTestCase(run_test_server.TestCaseWithServers): keep_client.get, unsigned_bar_locator) - -# KeepOptionalPermission: starts Keep with --permission-key-file -# but not --enforce-permissions (i.e. generate signatures on PUT -# requests, but do not require them for GET requests) -# -# All of these requests should succeed when permissions are optional: -# * authenticated request, signed locator -# * authenticated request, unsigned locator -# * unauthenticated request, signed locator -# * unauthenticated request, unsigned locator -class KeepOptionalPermission(run_test_server.TestCaseWithServers): - MAIN_SERVER = {} - KEEP_SERVER = {'blob_signing_key': 'abcdefghijk0123456789', - 'enforce_permissions': False} - - @classmethod - def setUpClass(cls): - super(KeepOptionalPermission, cls).setUpClass() - run_test_server.authorize_with("admin") - cls.api_client = arvados.api('v1') - - def setUp(self): - super(KeepOptionalPermission, self).setUp() - self.keep_client = arvados.KeepClient(api_client=self.api_client, - proxy='', local_store='') - - def _put_foo_and_check(self): - signed_locator = self.keep_client.put('foo') - self.assertRegexpMatches( - signed_locator, - r'^acbd18db4cc2f85cedef654fccc4a4d8\+3\+A[a-f0-9]+@[a-f0-9]+$', - 'invalid locator from Keep.put("foo"): ' + signed_locator) - return signed_locator - - def test_KeepAuthenticatedSignedTest(self): - signed_locator = self._put_foo_and_check() - self.assertEqual(self.keep_client.get(signed_locator), - 'foo', - 'wrong content from Keep.get(md5("foo"))') - - def test_KeepAuthenticatedUnsignedTest(self): - signed_locator = self._put_foo_and_check() - self.assertEqual(self.keep_client.get("acbd18db4cc2f85cedef654fccc4a4d8"), - 'foo', - 'wrong content from Keep.get(md5("foo"))') - - def test_KeepUnauthenticatedSignedTest(self): - # Check that signed GET requests work even when permissions - # enforcement is off. - signed_locator = self._put_foo_and_check() - self.keep_client.api_token = '' - self.assertEqual(self.keep_client.get(signed_locator), - 'foo', - 'wrong content from Keep.get(md5("foo"))') - - def test_KeepUnauthenticatedUnsignedTest(self): - # Since --enforce-permissions is not in effect, GET requests - # need not be authenticated. - signed_locator = self._put_foo_and_check() - self.keep_client.api_token = '' - self.assertEqual(self.keep_client.get("acbd18db4cc2f85cedef654fccc4a4d8"), - 'foo', - 'wrong content from Keep.get(md5("foo"))') - - -class KeepProxyTestCase(run_test_server.TestCaseWithServers): +@parameterized.parameterized_class([{"disk_cache": True}, {"disk_cache": False}]) +class KeepProxyTestCase(run_test_server.TestCaseWithServers, DiskCacheBase): + disk_cache = False MAIN_SERVER = {} KEEP_SERVER = {} KEEP_PROXY_SERVER = {} @@ -225,46 +205,53 @@ class KeepProxyTestCase(run_test_server.TestCaseWithServers): cls.api_client = arvados.api('v1') def tearDown(self): - arvados.config.settings().pop('ARVADOS_EXTERNAL_CLIENT', None) super(KeepProxyTestCase, self).tearDown() + DiskCacheBase.tearDown(self) def test_KeepProxyTest1(self): - # Will use ARVADOS_KEEP_PROXY environment variable that is set by - # setUpClass(). + # Will use ARVADOS_KEEP_SERVICES environment variable that + # is set by setUpClass(). keep_client = arvados.KeepClient(api_client=self.api_client, - local_store='') + local_store='', block_cache=self.make_block_cache(self.disk_cache)) baz_locator = keep_client.put('baz') - self.assertRegexpMatches( + self.assertRegex( baz_locator, '^73feffa4b7f6bb68e44cf984c85f6e88\+3', 'wrong md5 hash from Keep.put("baz"): ' + baz_locator) self.assertEqual(keep_client.get(baz_locator), - 'baz', + b'baz', 'wrong content from Keep.get(md5("baz"))') self.assertTrue(keep_client.using_proxy) - def test_KeepProxyTest2(self): - # Don't instantiate the proxy directly, but set the X-External-Client - # header. The API server should direct us to the proxy. - arvados.config.settings()['ARVADOS_EXTERNAL_CLIENT'] = 'true' + def test_KeepProxyTestMultipleURIs(self): + # Test using ARVADOS_KEEP_SERVICES env var overriding any + # existing proxy setting and setting multiple proxies + arvados.config.settings()['ARVADOS_KEEP_SERVICES'] = 'http://10.0.0.1 https://foo.example.org:1234/' keep_client = arvados.KeepClient(api_client=self.api_client, - proxy='', local_store='') - baz_locator = keep_client.put('baz2') - self.assertRegexpMatches( - baz_locator, - '^91f372a266fe2bf2823cb8ec7fda31ce\+4', - 'wrong md5 hash from Keep.put("baz2"): ' + baz_locator) - self.assertEqual(keep_client.get(baz_locator), - 'baz2', - 'wrong content from Keep.get(md5("baz2"))') - self.assertTrue(keep_client.using_proxy) + local_store='', + block_cache=self.make_block_cache(self.disk_cache)) + uris = [x['_service_root'] for x in keep_client._keep_services] + self.assertEqual(uris, ['http://10.0.0.1/', + 'https://foo.example.org:1234/']) + + def test_KeepProxyTestInvalidURI(self): + arvados.config.settings()['ARVADOS_KEEP_SERVICES'] = 'bad.uri.org' + with self.assertRaises(arvados.errors.ArgumentError): + keep_client = arvados.KeepClient(api_client=self.api_client, + local_store='', + block_cache=self.make_block_cache(self.disk_cache)) +@parameterized.parameterized_class([{"disk_cache": True}, {"disk_cache": False}]) +class KeepClientServiceTestCase(unittest.TestCase, tutil.ApiClientMock, DiskCacheBase): + disk_cache = False + + def tearDown(self): + DiskCacheBase.tearDown(self) -class KeepClientServiceTestCase(unittest.TestCase, tutil.ApiClientMock): def get_service_roots(self, api_client): - keep_client = arvados.KeepClient(api_client=api_client) + keep_client = arvados.KeepClient(api_client=api_client, block_cache=self.make_block_cache(self.disk_cache)) services = keep_client.weighted_service_roots(arvados.KeepLocator('0'*32)) - return [urlparse.urlparse(url) for url in sorted(services)] + return [urllib.parse.urlparse(url) for url in sorted(services)] def test_ssl_flag_respected_in_roots(self): for ssl_flag in [False, True]: @@ -279,6 +266,66 @@ class KeepClientServiceTestCase(unittest.TestCase, tutil.ApiClientMock): self.assertEqual('100::1', service.hostname) self.assertEqual(10, service.port) + def test_recognize_proxy_services_in_controller_response(self): + keep_client = arvados.KeepClient(api_client=self.mock_keep_services( + service_type='proxy', service_host='localhost', service_port=9, count=1), + block_cache=self.make_block_cache(self.disk_cache)) + try: + # this will fail, but it ensures we get the service + # discovery response + keep_client.put('baz2') + except: + pass + self.assertTrue(keep_client.using_proxy) + + def test_insecure_disables_tls_verify(self): + api_client = self.mock_keep_services(count=1) + force_timeout = socket.timeout("timed out") + + api_client.insecure = True + with tutil.mock_keep_responses(b'foo', 200) as mock: + keep_client = arvados.KeepClient(api_client=api_client, block_cache=self.make_block_cache(self.disk_cache)) + keep_client.get('acbd18db4cc2f85cedef654fccc4a4d8+3') + self.assertEqual( + mock.responses[0].getopt(pycurl.SSL_VERIFYPEER), + 0) + self.assertEqual( + mock.responses[0].getopt(pycurl.SSL_VERIFYHOST), + 0) + + api_client.insecure = False + with tutil.mock_keep_responses(b'foo', 200) as mock: + keep_client = arvados.KeepClient(api_client=api_client, block_cache=self.make_block_cache(self.disk_cache)) + keep_client.get('acbd18db4cc2f85cedef654fccc4a4d8+3') + # getopt()==None here means we didn't change the + # default. If we were using real pycurl instead of a mock, + # it would return the default value 1. + self.assertEqual( + mock.responses[0].getopt(pycurl.SSL_VERIFYPEER), + None) + self.assertEqual( + mock.responses[0].getopt(pycurl.SSL_VERIFYHOST), + None) + + def test_refresh_signature(self): + blk_digest = '6f5902ac237024bdd0c176cb93063dc4+11' + blk_sig = 'da39a3ee5e6b4b0d3255bfef95601890afd80709@53bed294' + local_loc = blk_digest+'+A'+blk_sig + remote_loc = blk_digest+'+R'+blk_sig + api_client = self.mock_keep_services(count=1) + headers = {'X-Keep-Locator':local_loc} + with tutil.mock_keep_responses('', 200, **headers): + # Check that the translated locator gets returned + keep_client = arvados.KeepClient(api_client=api_client, block_cache=self.make_block_cache(self.disk_cache)) + self.assertEqual(local_loc, keep_client.refresh_signature(remote_loc)) + # Check that refresh_signature() uses the correct method and headers + keep_client._get_or_head = mock.MagicMock() + keep_client.refresh_signature(remote_loc) + args, kwargs = keep_client._get_or_head.call_args_list[0] + self.assertIn(remote_loc, args) + self.assertEqual("HEAD", kwargs['method']) + self.assertIn('X-Keep-Signature', kwargs['headers']) + # test_*_timeout verify that KeepClient instructs pycurl to use # the appropriate connection and read timeouts. They don't care # whether pycurl actually exhibits the expected timeout behavior @@ -288,7 +335,7 @@ class KeepClientServiceTestCase(unittest.TestCase, tutil.ApiClientMock): api_client = self.mock_keep_services(count=1) force_timeout = socket.timeout("timed out") with tutil.mock_keep_responses(force_timeout, 0) as mock: - keep_client = arvados.KeepClient(api_client=api_client) + keep_client = arvados.KeepClient(api_client=api_client, block_cache=self.make_block_cache(self.disk_cache)) with self.assertRaises(arvados.errors.KeepReadError): keep_client.get('ffffffffffffffffffffffffffffffff') self.assertEqual( @@ -305,9 +352,9 @@ class KeepClientServiceTestCase(unittest.TestCase, tutil.ApiClientMock): api_client = self.mock_keep_services(count=1) force_timeout = socket.timeout("timed out") with tutil.mock_keep_responses(force_timeout, 0) as mock: - keep_client = arvados.KeepClient(api_client=api_client) + keep_client = arvados.KeepClient(api_client=api_client, block_cache=self.make_block_cache(self.disk_cache)) with self.assertRaises(arvados.errors.KeepWriteError): - keep_client.put('foo') + keep_client.put(b'foo') self.assertEqual( mock.responses[0].getopt(pycurl.CONNECTTIMEOUT_MS), int(arvados.KeepClient.DEFAULT_TIMEOUT[0]*1000)) @@ -318,11 +365,28 @@ class KeepClientServiceTestCase(unittest.TestCase, tutil.ApiClientMock): mock.responses[0].getopt(pycurl.LOW_SPEED_LIMIT), int(arvados.KeepClient.DEFAULT_TIMEOUT[2])) + def test_head_timeout(self): + api_client = self.mock_keep_services(count=1) + force_timeout = socket.timeout("timed out") + with tutil.mock_keep_responses(force_timeout, 0) as mock: + keep_client = arvados.KeepClient(api_client=api_client, block_cache=self.make_block_cache(self.disk_cache)) + with self.assertRaises(arvados.errors.KeepReadError): + keep_client.head('ffffffffffffffffffffffffffffffff') + self.assertEqual( + mock.responses[0].getopt(pycurl.CONNECTTIMEOUT_MS), + int(arvados.KeepClient.DEFAULT_TIMEOUT[0]*1000)) + self.assertEqual( + mock.responses[0].getopt(pycurl.LOW_SPEED_TIME), + None) + self.assertEqual( + mock.responses[0].getopt(pycurl.LOW_SPEED_LIMIT), + None) + def test_proxy_get_timeout(self): api_client = self.mock_keep_services(service_type='proxy', count=1) force_timeout = socket.timeout("timed out") with tutil.mock_keep_responses(force_timeout, 0) as mock: - keep_client = arvados.KeepClient(api_client=api_client) + keep_client = arvados.KeepClient(api_client=api_client, block_cache=self.make_block_cache(self.disk_cache)) with self.assertRaises(arvados.errors.KeepReadError): keep_client.get('ffffffffffffffffffffffffffffffff') self.assertEqual( @@ -335,7 +399,25 @@ class KeepClientServiceTestCase(unittest.TestCase, tutil.ApiClientMock): mock.responses[0].getopt(pycurl.LOW_SPEED_LIMIT), int(arvados.KeepClient.DEFAULT_PROXY_TIMEOUT[2])) + def test_proxy_head_timeout(self): + api_client = self.mock_keep_services(service_type='proxy', count=1) + force_timeout = socket.timeout("timed out") + with tutil.mock_keep_responses(force_timeout, 0) as mock: + keep_client = arvados.KeepClient(api_client=api_client, block_cache=self.make_block_cache(self.disk_cache)) + with self.assertRaises(arvados.errors.KeepReadError): + keep_client.head('ffffffffffffffffffffffffffffffff') + self.assertEqual( + mock.responses[0].getopt(pycurl.CONNECTTIMEOUT_MS), + int(arvados.KeepClient.DEFAULT_PROXY_TIMEOUT[0]*1000)) + self.assertEqual( + mock.responses[0].getopt(pycurl.LOW_SPEED_TIME), + None) + self.assertEqual( + mock.responses[0].getopt(pycurl.LOW_SPEED_LIMIT), + None) + def test_proxy_put_timeout(self): + self.disk_cache_dir = None api_client = self.mock_keep_services(service_type='proxy', count=1) force_timeout = socket.timeout("timed out") with tutil.mock_keep_responses(force_timeout, 0) as mock: @@ -356,7 +438,7 @@ class KeepClientServiceTestCase(unittest.TestCase, tutil.ApiClientMock): api_client = mock.MagicMock(name='api_client') api_client.keep_services().accessible().execute.side_effect = ( arvados.errors.ApiError) - keep_client = arvados.KeepClient(api_client=api_client) + keep_client = arvados.KeepClient(api_client=api_client, block_cache=self.make_block_cache(self.disk_cache)) with self.assertRaises(exc_class) as err_check: getattr(keep_client, verb)('d41d8cd98f00b204e9800998ecf8427e+0') self.assertEqual(0, len(err_check.exception.request_errors())) @@ -364,25 +446,32 @@ class KeepClientServiceTestCase(unittest.TestCase, tutil.ApiClientMock): def test_get_error_with_no_services(self): self.check_no_services_error('get', arvados.errors.KeepReadError) + def test_head_error_with_no_services(self): + self.check_no_services_error('head', arvados.errors.KeepReadError) + def test_put_error_with_no_services(self): self.check_no_services_error('put', arvados.errors.KeepWriteError) def check_errors_from_last_retry(self, verb, exc_class): api_client = self.mock_keep_services(count=2) req_mock = tutil.mock_keep_responses( - "retry error reporting test", 500, 500, 403, 403) + "retry error reporting test", 500, 500, 500, 500, 500, 500, 502, 502) with req_mock, tutil.skip_sleep, \ self.assertRaises(exc_class) as err_check: - keep_client = arvados.KeepClient(api_client=api_client) + keep_client = arvados.KeepClient(api_client=api_client, block_cache=self.make_block_cache(self.disk_cache)) getattr(keep_client, verb)('d41d8cd98f00b204e9800998ecf8427e+0', num_retries=3) - self.assertEqual([403, 403], [ + self.assertEqual([502, 502], [ getattr(error, 'status_code', None) - for error in err_check.exception.request_errors().itervalues()]) + for error in err_check.exception.request_errors().values()]) + self.assertRegex(str(err_check.exception), r'failed to (read|write) .* after 4 attempts') def test_get_error_reflects_last_retry(self): self.check_errors_from_last_retry('get', arvados.errors.KeepReadError) + def test_head_error_reflects_last_retry(self): + self.check_errors_from_last_retry('head', arvados.errors.KeepReadError) + def test_put_error_reflects_last_retry(self): self.check_errors_from_last_retry('put', arvados.errors.KeepWriteError) @@ -392,7 +481,7 @@ class KeepClientServiceTestCase(unittest.TestCase, tutil.ApiClientMock): api_client = self.mock_keep_services(count=3) with tutil.mock_keep_responses(data_loc, 200, 500, 500) as req_mock, \ self.assertRaises(arvados.errors.KeepWriteError) as exc_check: - keep_client = arvados.KeepClient(api_client=api_client) + keep_client = arvados.KeepClient(api_client=api_client, block_cache=self.make_block_cache(self.disk_cache)) keep_client.put(data) self.assertEqual(2, len(exc_check.exception.request_errors())) @@ -402,44 +491,298 @@ class KeepClientServiceTestCase(unittest.TestCase, tutil.ApiClientMock): api_client = self.mock_keep_services(service_type='proxy', read_only=True, count=1) with tutil.mock_keep_responses(data_loc, 200, 500, 500) as req_mock, \ self.assertRaises(arvados.errors.KeepWriteError) as exc_check: - keep_client = arvados.KeepClient(api_client=api_client) + keep_client = arvados.KeepClient(api_client=api_client, block_cache=self.make_block_cache(self.disk_cache)) keep_client.put(data) self.assertEqual(True, ("no Keep services available" in str(exc_check.exception))) self.assertEqual(0, len(exc_check.exception.request_errors())) def test_oddball_service_get(self): - body = 'oddball service get' + body = b'oddball service get' api_client = self.mock_keep_services(service_type='fancynewblobstore') with tutil.mock_keep_responses(body, 200): - keep_client = arvados.KeepClient(api_client=api_client) + keep_client = arvados.KeepClient(api_client=api_client, block_cache=self.make_block_cache(self.disk_cache)) actual = keep_client.get(tutil.str_keep_locator(body)) self.assertEqual(body, actual) def test_oddball_service_put(self): - body = 'oddball service put' + body = b'oddball service put' pdh = tutil.str_keep_locator(body) api_client = self.mock_keep_services(service_type='fancynewblobstore') with tutil.mock_keep_responses(pdh, 200): - keep_client = arvados.KeepClient(api_client=api_client) + keep_client = arvados.KeepClient(api_client=api_client, block_cache=self.make_block_cache(self.disk_cache)) actual = keep_client.put(body, copies=1) self.assertEqual(pdh, actual) def test_oddball_service_writer_count(self): - body = 'oddball service writer count' + body = b'oddball service writer count' pdh = tutil.str_keep_locator(body) api_client = self.mock_keep_services(service_type='fancynewblobstore', count=4) headers = {'x-keep-replicas-stored': 3} with tutil.mock_keep_responses(pdh, 200, 418, 418, 418, **headers) as req_mock: - keep_client = arvados.KeepClient(api_client=api_client) + keep_client = arvados.KeepClient(api_client=api_client, block_cache=self.make_block_cache(self.disk_cache)) actual = keep_client.put(body, copies=2) self.assertEqual(pdh, actual) self.assertEqual(1, req_mock.call_count) +@tutil.skip_sleep +@parameterized.parameterized_class([{"disk_cache": True}, {"disk_cache": False}]) +class KeepClientCacheTestCase(unittest.TestCase, tutil.ApiClientMock, DiskCacheBase): + disk_cache = False + + def setUp(self): + self.api_client = self.mock_keep_services(count=2) + self.keep_client = arvados.KeepClient(api_client=self.api_client, block_cache=self.make_block_cache(self.disk_cache)) + self.data = b'xyzzy' + self.locator = '1271ed5ef305aadabc605b1609e24c52' + + def tearDown(self): + DiskCacheBase.tearDown(self) + + @mock.patch('arvados.KeepClient.KeepService.get') + def test_get_request_cache(self, get_mock): + with tutil.mock_keep_responses(self.data, 200, 200): + self.keep_client.get(self.locator) + self.keep_client.get(self.locator) + # Request already cached, don't require more than one request + get_mock.assert_called_once() + + @mock.patch('arvados.KeepClient.KeepService.get') + def test_head_request_cache(self, get_mock): + with tutil.mock_keep_responses(self.data, 200, 200): + self.keep_client.head(self.locator) + self.keep_client.head(self.locator) + # Don't cache HEAD requests so that they're not confused with GET reqs + self.assertEqual(2, get_mock.call_count) + + @mock.patch('arvados.KeepClient.KeepService.get') + def test_head_and_then_get_return_different_responses(self, get_mock): + head_resp = None + get_resp = None + get_mock.side_effect = [b'first response', b'second response'] + with tutil.mock_keep_responses(self.data, 200, 200): + head_resp = self.keep_client.head(self.locator) + get_resp = self.keep_client.get(self.locator) + self.assertEqual(b'first response', head_resp) + # First reponse was not cached because it was from a HEAD request. + self.assertNotEqual(head_resp, get_resp) + +@tutil.skip_sleep +@parameterized.parameterized_class([{"disk_cache": True}, {"disk_cache": False}]) +class KeepStorageClassesTestCase(unittest.TestCase, tutil.ApiClientMock, DiskCacheBase): + disk_cache = False + + def setUp(self): + self.api_client = self.mock_keep_services(count=2) + self.keep_client = arvados.KeepClient(api_client=self.api_client, block_cache=self.make_block_cache(self.disk_cache)) + self.data = b'xyzzy' + self.locator = '1271ed5ef305aadabc605b1609e24c52' + + def tearDown(self): + DiskCacheBase.tearDown(self) + + def test_multiple_default_storage_classes_req_header(self): + api_mock = self.api_client_mock() + api_mock.config.return_value = { + 'StorageClasses': { + 'foo': { 'Default': True }, + 'bar': { 'Default': True }, + 'baz': { 'Default': False } + } + } + api_client = self.mock_keep_services(api_mock=api_mock, count=2) + keep_client = arvados.KeepClient(api_client=api_client, block_cache=self.make_block_cache(self.disk_cache)) + resp_hdr = { + 'x-keep-storage-classes-confirmed': 'foo=1, bar=1', + 'x-keep-replicas-stored': 1 + } + with tutil.mock_keep_responses(self.locator, 200, **resp_hdr) as mock: + keep_client.put(self.data, copies=1) + req_hdr = mock.responses[0] + self.assertIn( + 'X-Keep-Storage-Classes: bar, foo', req_hdr.getopt(pycurl.HTTPHEADER)) + + def test_storage_classes_req_header(self): + self.assertEqual( + self.api_client.config()['StorageClasses'], + {'default': {'Default': True}}) + cases = [ + # requested, expected + [['foo'], 'X-Keep-Storage-Classes: foo'], + [['bar', 'foo'], 'X-Keep-Storage-Classes: bar, foo'], + [[], 'X-Keep-Storage-Classes: default'], + [None, 'X-Keep-Storage-Classes: default'], + ] + for req_classes, expected_header in cases: + headers = {'x-keep-replicas-stored': 1} + if req_classes is None or len(req_classes) == 0: + confirmed_hdr = 'default=1' + elif len(req_classes) > 0: + confirmed_hdr = ', '.join(["{}=1".format(cls) for cls in req_classes]) + headers.update({'x-keep-storage-classes-confirmed': confirmed_hdr}) + with tutil.mock_keep_responses(self.locator, 200, **headers) as mock: + self.keep_client.put(self.data, copies=1, classes=req_classes) + req_hdr = mock.responses[0] + self.assertIn(expected_header, req_hdr.getopt(pycurl.HTTPHEADER)) + + def test_partial_storage_classes_put(self): + headers = { + 'x-keep-replicas-stored': 1, + 'x-keep-storage-classes-confirmed': 'foo=1'} + with tutil.mock_keep_responses(self.locator, 200, 503, **headers) as mock: + with self.assertRaises(arvados.errors.KeepWriteError): + self.keep_client.put(self.data, copies=1, classes=['foo', 'bar']) + # 1st request, both classes pending + req1_headers = mock.responses[0].getopt(pycurl.HTTPHEADER) + self.assertIn('X-Keep-Storage-Classes: bar, foo', req1_headers) + # 2nd try, 'foo' class already satisfied + req2_headers = mock.responses[1].getopt(pycurl.HTTPHEADER) + self.assertIn('X-Keep-Storage-Classes: bar', req2_headers) + + def test_successful_storage_classes_put_requests(self): + cases = [ + # wanted_copies, wanted_classes, confirmed_copies, confirmed_classes, expected_requests + [ 1, ['foo'], 1, 'foo=1', 1], + [ 1, ['foo'], 2, 'foo=2', 1], + [ 2, ['foo'], 2, 'foo=2', 1], + [ 2, ['foo'], 1, 'foo=1', 2], + [ 1, ['foo', 'bar'], 1, 'foo=1, bar=1', 1], + [ 1, ['foo', 'bar'], 2, 'foo=2, bar=2', 1], + [ 2, ['foo', 'bar'], 2, 'foo=2, bar=2', 1], + [ 2, ['foo', 'bar'], 1, 'foo=1, bar=1', 2], + [ 1, ['foo', 'bar'], 1, None, 1], + [ 1, ['foo'], 1, None, 1], + [ 2, ['foo'], 2, None, 1], + [ 2, ['foo'], 1, None, 2], + ] + for w_copies, w_classes, c_copies, c_classes, e_reqs in cases: + headers = {'x-keep-replicas-stored': c_copies} + if c_classes is not None: + headers.update({'x-keep-storage-classes-confirmed': c_classes}) + with tutil.mock_keep_responses(self.locator, 200, 200, **headers) as mock: + case_desc = 'wanted_copies={}, wanted_classes="{}", confirmed_copies={}, confirmed_classes="{}", expected_requests={}'.format(w_copies, ', '.join(w_classes), c_copies, c_classes, e_reqs) + self.assertEqual(self.locator, + self.keep_client.put(self.data, copies=w_copies, classes=w_classes), + case_desc) + self.assertEqual(e_reqs, mock.call_count, case_desc) + + def test_failed_storage_classes_put_requests(self): + cases = [ + # wanted_copies, wanted_classes, confirmed_copies, confirmed_classes, return_code + [ 1, ['foo'], 1, 'bar=1', 200], + [ 1, ['foo'], 1, None, 503], + [ 2, ['foo'], 1, 'bar=1, foo=0', 200], + [ 3, ['foo'], 1, 'bar=1, foo=1', 200], + [ 3, ['foo', 'bar'], 1, 'bar=2, foo=1', 200], + ] + for w_copies, w_classes, c_copies, c_classes, return_code in cases: + headers = {'x-keep-replicas-stored': c_copies} + if c_classes is not None: + headers.update({'x-keep-storage-classes-confirmed': c_classes}) + with tutil.mock_keep_responses(self.locator, return_code, return_code, **headers): + case_desc = 'wanted_copies={}, wanted_classes="{}", confirmed_copies={}, confirmed_classes="{}"'.format(w_copies, ', '.join(w_classes), c_copies, c_classes) + with self.assertRaises(arvados.errors.KeepWriteError, msg=case_desc): + self.keep_client.put(self.data, copies=w_copies, classes=w_classes) + +@tutil.skip_sleep +@parameterized.parameterized_class([{"disk_cache": True}, {"disk_cache": False}]) +class KeepXRequestIdTestCase(unittest.TestCase, tutil.ApiClientMock, DiskCacheBase): + disk_cache = False + + def setUp(self): + self.api_client = self.mock_keep_services(count=2) + self.keep_client = arvados.KeepClient(api_client=self.api_client, block_cache=self.make_block_cache(self.disk_cache)) + self.data = b'xyzzy' + self.locator = '1271ed5ef305aadabc605b1609e24c52' + self.test_id = arvados.util.new_request_id() + self.assertRegex(self.test_id, r'^req-[a-z0-9]{20}$') + # If we don't set request_id to None explicitly here, it will + # return : + self.api_client.request_id = None + + def tearDown(self): + DiskCacheBase.tearDown(self) + + def test_default_to_api_client_request_id(self): + self.api_client.request_id = self.test_id + with tutil.mock_keep_responses(self.locator, 200, 200) as mock: + self.keep_client.put(self.data) + self.assertEqual(2, len(mock.responses)) + for resp in mock.responses: + self.assertProvidedRequestId(resp) + + with tutil.mock_keep_responses(self.data, 200) as mock: + self.keep_client.get(self.locator) + self.assertProvidedRequestId(mock.responses[0]) + + with tutil.mock_keep_responses(b'', 200) as mock: + self.keep_client.head(self.locator) + self.assertProvidedRequestId(mock.responses[0]) + + def test_explicit_request_id(self): + with tutil.mock_keep_responses(self.locator, 200, 200) as mock: + self.keep_client.put(self.data, request_id=self.test_id) + self.assertEqual(2, len(mock.responses)) + for resp in mock.responses: + self.assertProvidedRequestId(resp) + + with tutil.mock_keep_responses(self.data, 200) as mock: + self.keep_client.get(self.locator, request_id=self.test_id) + self.assertProvidedRequestId(mock.responses[0]) + + with tutil.mock_keep_responses(b'', 200) as mock: + self.keep_client.head(self.locator, request_id=self.test_id) + self.assertProvidedRequestId(mock.responses[0]) + + def test_automatic_request_id(self): + with tutil.mock_keep_responses(self.locator, 200, 200) as mock: + self.keep_client.put(self.data) + self.assertEqual(2, len(mock.responses)) + for resp in mock.responses: + self.assertAutomaticRequestId(resp) + + with tutil.mock_keep_responses(self.data, 200) as mock: + self.keep_client.get(self.locator) + self.assertAutomaticRequestId(mock.responses[0]) + + with tutil.mock_keep_responses(b'', 200) as mock: + self.keep_client.head(self.locator) + self.assertAutomaticRequestId(mock.responses[0]) + + def test_request_id_in_exception(self): + with tutil.mock_keep_responses(b'', 400, 400, 400) as mock: + with self.assertRaisesRegex(arvados.errors.KeepReadError, self.test_id): + self.keep_client.head(self.locator, request_id=self.test_id) + + with tutil.mock_keep_responses(b'', 400, 400, 400) as mock: + with self.assertRaisesRegex(arvados.errors.KeepReadError, r'req-[a-z0-9]{20}'): + self.keep_client.get(self.locator) + + with tutil.mock_keep_responses(b'', 400, 400, 400) as mock: + with self.assertRaisesRegex(arvados.errors.KeepWriteError, self.test_id): + self.keep_client.put(self.data, request_id=self.test_id) + + with tutil.mock_keep_responses(b'', 400, 400, 400) as mock: + with self.assertRaisesRegex(arvados.errors.KeepWriteError, r'req-[a-z0-9]{20}'): + self.keep_client.put(self.data) + + def assertAutomaticRequestId(self, resp): + hdr = [x for x in resp.getopt(pycurl.HTTPHEADER) + if x.startswith('X-Request-Id: ')][0] + self.assertNotEqual(hdr, 'X-Request-Id: '+self.test_id) + self.assertRegex(hdr, r'^X-Request-Id: req-[a-z0-9]{20}$') + + def assertProvidedRequestId(self, resp): + self.assertIn('X-Request-Id: '+self.test_id, + resp.getopt(pycurl.HTTPHEADER)) + @tutil.skip_sleep -class KeepClientRendezvousTestCase(unittest.TestCase, tutil.ApiClientMock): +#@parameterized.parameterized_class([{"disk_cache": True}, {"disk_cache": False}]) +class KeepClientRendezvousTestCase(unittest.TestCase, tutil.ApiClientMock, DiskCacheBase): + disk_cache = False def setUp(self): # expected_order[i] is the probe order for @@ -456,13 +799,16 @@ class KeepClientRendezvousTestCase(unittest.TestCase, tutil.ApiClientMock): list('9d81c02e76a3bf54'), ] self.blocks = [ - "{:064x}".format(x) + "{:064x}".format(x).encode() for x in range(len(self.expected_order))] self.hashes = [ hashlib.md5(self.blocks[x]).hexdigest() for x in range(len(self.expected_order))] self.api_client = self.mock_keep_services(count=self.services) - self.keep_client = arvados.KeepClient(api_client=self.api_client) + self.keep_client = arvados.KeepClient(api_client=self.api_client, block_cache=self.make_block_cache(self.disk_cache)) + + def tearDown(self): + DiskCacheBase.tearDown(self) def test_weighted_service_roots_against_reference_set(self): # Confirm weighted_service_roots() returns the correct order @@ -477,6 +823,10 @@ class KeepClientRendezvousTestCase(unittest.TestCase, tutil.ApiClientMock): self._test_probe_order_against_reference_set( lambda i: self.keep_client.get(self.hashes[i], num_retries=1)) + def test_head_probe_order_against_reference_set(self): + self._test_probe_order_against_reference_set( + lambda i: self.keep_client.head(self.hashes[i], num_retries=1)) + def test_put_probe_order_against_reference_set(self): # copies=1 prevents the test from being sensitive to races # between writer threads. @@ -489,7 +839,7 @@ class KeepClientRendezvousTestCase(unittest.TestCase, tutil.ApiClientMock): self.assertRaises(arvados.errors.KeepRequestError): op(i) got_order = [ - re.search(r'//\[?keep0x([0-9a-f]+)', resp.getopt(pycurl.URL)).group(1) + re.search(r'//\[?keep0x([0-9a-f]+)', resp.getopt(pycurl.URL).decode()).group(1) for resp in mock.responses] self.assertEqual(self.expected_order[i]*2, got_order) @@ -500,7 +850,7 @@ class KeepClientRendezvousTestCase(unittest.TestCase, tutil.ApiClientMock): self.assertRaises(arvados.errors.KeepWriteError): self.keep_client.put(self.blocks[i], num_retries=2, copies=copies) got_order = [ - re.search(r'//\[?keep0x([0-9a-f]+)', resp.getopt(pycurl.URL)).group(1) + re.search(r'//\[?keep0x([0-9a-f]+)', resp.getopt(pycurl.URL).decode()).group(1) for resp in mock.responses] # With T threads racing to make requests, the position # of a given server in the sequence of HTTP requests @@ -528,15 +878,15 @@ class KeepClientRendezvousTestCase(unittest.TestCase, tutil.ApiClientMock): def test_probe_waste_adding_one_server(self): hashes = [ - hashlib.md5("{:064x}".format(x)).hexdigest() for x in range(100)] + hashlib.md5("{:064x}".format(x).encode()).hexdigest() for x in range(100)] initial_services = 12 self.api_client = self.mock_keep_services(count=initial_services) - self.keep_client = arvados.KeepClient(api_client=self.api_client) + self.keep_client = arvados.KeepClient(api_client=self.api_client, block_cache=self.make_block_cache(self.disk_cache)) probes_before = [ self.keep_client.weighted_service_roots(arvados.KeepLocator(hash)) for hash in hashes] for added_services in range(1, 12): api_client = self.mock_keep_services(count=initial_services+added_services) - keep_client = arvados.KeepClient(api_client=api_client) + keep_client = arvados.KeepClient(api_client=api_client, block_cache=self.make_block_cache(self.disk_cache)) total_penalty = 0 for hash_index in range(len(hashes)): probe_after = keep_client.weighted_service_roots( @@ -566,18 +916,18 @@ class KeepClientRendezvousTestCase(unittest.TestCase, tutil.ApiClientMock): max_penalty)) def check_64_zeros_error_order(self, verb, exc_class): - data = '0' * 64 + data = b'0' * 64 if verb == 'get': data = tutil.str_keep_locator(data) # Arbitrary port number: aport = random.randint(1024,65535) api_client = self.mock_keep_services(service_port=aport, count=self.services) - keep_client = arvados.KeepClient(api_client=api_client) + keep_client = arvados.KeepClient(api_client=api_client, block_cache=self.make_block_cache(self.disk_cache)) with mock.patch('pycurl.Curl') as curl_mock, \ self.assertRaises(exc_class) as err_check: - curl_mock.return_value.side_effect = socket.timeout + curl_mock.return_value = tutil.FakeCurl.make(code=500, body=b'') getattr(keep_client, verb)(data) - urls = [urlparse.urlparse(url) + urls = [urllib.parse.urlparse(url) for url in err_check.exception.request_errors()] self.assertEqual([('keep0x' + c, aport) for c in '3eab2d5fc9681074'], [(url.hostname, url.port) for url in urls]) @@ -588,15 +938,20 @@ class KeepClientRendezvousTestCase(unittest.TestCase, tutil.ApiClientMock): def test_put_error_shows_probe_order(self): self.check_64_zeros_error_order('put', arvados.errors.KeepWriteError) +@parameterized.parameterized_class([{"disk_cache": True}, {"disk_cache": False}]) +class KeepClientTimeout(keepstub.StubKeepServers, unittest.TestCase, DiskCacheBase): + disk_cache = False -class KeepClientTimeout(unittest.TestCase, tutil.ApiClientMock): # BANDWIDTH_LOW_LIM must be less than len(DATA) so we can transfer # 1s worth of data and then trigger bandwidth errors before running # out of data. - DATA = 'x'*2**11 + DATA = b'x'*2**11 BANDWIDTH_LOW_LIM = 1024 TIMEOUT_TIME = 1.0 + def tearDown(self): + DiskCacheBase.tearDown(self) + class assertTakesBetween(unittest.TestCase): def __init__(self, tmin, tmax): self.tmin = tmin @@ -623,28 +978,10 @@ class KeepClientTimeout(unittest.TestCase, tutil.ApiClientMock): delta = round(time.time() - self.t0, 3) self.assertGreaterEqual(delta, self.tmin) - def setUp(self): - sock = socket.socket() - sock.bind(('0.0.0.0', 0)) - self.port = sock.getsockname()[1] - sock.close() - self.server = keepstub.Server(('0.0.0.0', self.port), keepstub.Handler) - self.thread = threading.Thread(target=self.server.serve_forever) - self.thread.daemon = True # Exit thread if main proc exits - self.thread.start() - self.api_client = self.mock_keep_services( - count=1, - service_host='localhost', - service_port=self.port, - ) - - def tearDown(self): - self.server.shutdown() - def keepClient(self, timeouts=(0.1, TIMEOUT_TIME, BANDWIDTH_LOW_LIM)): return arvados.KeepClient( api_client=self.api_client, - timeout=timeouts) + timeout=timeouts, block_cache=self.make_block_cache(self.disk_cache)) def test_timeout_slow_connect(self): # Can't simulate TCP delays with our own socket. Leave our @@ -670,7 +1007,7 @@ class KeepClientTimeout(unittest.TestCase, tutil.ApiClientMock): loc = kc.put(self.DATA, copies=1, num_retries=0) self.server.setbandwidth(0.5*self.BANDWIDTH_LOW_LIM) with self.assertTakesGreater(self.TIMEOUT_TIME): - with self.assertRaises(arvados.errors.KeepReadError) as e: + with self.assertRaises(arvados.errors.KeepReadError): kc.get(loc, num_retries=0) with self.assertTakesGreater(self.TIMEOUT_TIME): with self.assertRaises(arvados.errors.KeepWriteError): @@ -680,19 +1017,25 @@ class KeepClientTimeout(unittest.TestCase, tutil.ApiClientMock): kc = self.keepClient() loc = kc.put(self.DATA, copies=1, num_retries=0) self.server.setbandwidth(self.BANDWIDTH_LOW_LIM) - self.server.setdelays(response=self.TIMEOUT_TIME) + # Note the actual delay must be 1s longer than the low speed + # limit interval in order for curl to detect it reliably. + self.server.setdelays(response=self.TIMEOUT_TIME+1) with self.assertTakesGreater(self.TIMEOUT_TIME): - with self.assertRaises(arvados.errors.KeepReadError) as e: + with self.assertRaises(arvados.errors.KeepReadError): kc.get(loc, num_retries=0) with self.assertTakesGreater(self.TIMEOUT_TIME): with self.assertRaises(arvados.errors.KeepWriteError): kc.put(self.DATA, copies=1, num_retries=0) + with self.assertTakesGreater(self.TIMEOUT_TIME): + kc.head(loc, num_retries=0) def test_low_bandwidth_with_server_mid_delay_failure(self): kc = self.keepClient() loc = kc.put(self.DATA, copies=1, num_retries=0) self.server.setbandwidth(self.BANDWIDTH_LOW_LIM) - self.server.setdelays(mid_write=self.TIMEOUT_TIME, mid_read=self.TIMEOUT_TIME) + # Note the actual delay must be 1s longer than the low speed + # limit interval in order for curl to detect it reliably. + self.server.setdelays(mid_write=self.TIMEOUT_TIME+1, mid_read=self.TIMEOUT_TIME+1) with self.assertTakesGreater(self.TIMEOUT_TIME): with self.assertRaises(arvados.errors.KeepReadError) as e: kc.get(loc, num_retries=0) @@ -735,15 +1078,20 @@ class KeepClientTimeout(unittest.TestCase, tutil.ApiClientMock): # Allow 10s to connect, then 1s for response. Nothing should # work, and everything should take at least 1s to return. kc = self.keepClient(timeouts=(10, 1)) - with self.assertTakesBetween(1, 1.9): + with self.assertTakesBetween(1, 9): with self.assertRaises(arvados.errors.KeepReadError): kc.get(loc, num_retries=0) - with self.assertTakesBetween(1, 1.9): + with self.assertTakesBetween(1, 9): with self.assertRaises(arvados.errors.KeepWriteError): kc.put(self.DATA, copies=1, num_retries=0) +@parameterized.parameterized_class([{"disk_cache": True}, {"disk_cache": False}]) +class KeepClientGatewayTestCase(unittest.TestCase, tutil.ApiClientMock, DiskCacheBase): + disk_cache = False + + def tearDown(self): + DiskCacheBase.tearDown(self) -class KeepClientGatewayTestCase(unittest.TestCase, tutil.ApiClientMock): def mock_disks_and_gateways(self, disks=3, gateways=1): self.gateways = [{ 'uuid': 'zzzzz-bi6l4-gateway{:08d}'.format(i), @@ -758,7 +1106,7 @@ class KeepClientGatewayTestCase(unittest.TestCase, tutil.ApiClientMock): for gw in self.gateways] self.api_client = self.mock_keep_services( count=disks, additional_services=self.gateways) - self.keepClient = arvados.KeepClient(api_client=self.api_client) + self.keepClient = arvados.KeepClient(api_client=self.api_client, block_cache=self.make_block_cache(self.disk_cache)) @mock.patch('pycurl.Curl') def test_get_with_gateway_hint_first(self, MockCurl): @@ -766,9 +1114,10 @@ class KeepClientGatewayTestCase(unittest.TestCase, tutil.ApiClientMock): code=200, body='foo', headers={'Content-Length': 3}) self.mock_disks_and_gateways() locator = 'acbd18db4cc2f85cedef654fccc4a4d8+3+K@' + self.gateways[0]['uuid'] - self.assertEqual('foo', self.keepClient.get(locator)) + self.assertEqual(b'foo', self.keepClient.get(locator)) self.assertEqual(self.gateway_roots[0]+locator, - MockCurl.return_value.getopt(pycurl.URL)) + MockCurl.return_value.getopt(pycurl.URL).decode()) + self.assertEqual(True, self.keepClient.head(locator)) @mock.patch('pycurl.Curl') def test_get_with_gateway_hints_in_order(self, MockCurl): @@ -787,25 +1136,60 @@ class KeepClientGatewayTestCase(unittest.TestCase, tutil.ApiClientMock): # Gateways are tried first, in the order given. for i, root in enumerate(self.gateway_roots): self.assertEqual(root+locator, - mocks[i].getopt(pycurl.URL)) + mocks[i].getopt(pycurl.URL).decode()) # Disk services are tried next. for i in range(gateways, gateways+disks): - self.assertRegexpMatches( - mocks[i].getopt(pycurl.URL), + self.assertRegex( + mocks[i].getopt(pycurl.URL).decode(), + r'keep0x') + + @mock.patch('pycurl.Curl') + def test_head_with_gateway_hints_in_order(self, MockCurl): + gateways = 4 + disks = 3 + mocks = [ + tutil.FakeCurl.make(code=404, body=b'') + for _ in range(gateways+disks) + ] + MockCurl.side_effect = tutil.queue_with(mocks) + self.mock_disks_and_gateways(gateways=gateways, disks=disks) + locator = '+'.join(['acbd18db4cc2f85cedef654fccc4a4d8+3'] + + ['K@'+gw['uuid'] for gw in self.gateways]) + with self.assertRaises(arvados.errors.NotFoundError): + self.keepClient.head(locator) + # Gateways are tried first, in the order given. + for i, root in enumerate(self.gateway_roots): + self.assertEqual(root+locator, + mocks[i].getopt(pycurl.URL).decode()) + # Disk services are tried next. + for i in range(gateways, gateways+disks): + self.assertRegex( + mocks[i].getopt(pycurl.URL).decode(), r'keep0x') @mock.patch('pycurl.Curl') def test_get_with_remote_proxy_hint(self, MockCurl): MockCurl.return_value = tutil.FakeCurl.make( - code=200, body='foo', headers={'Content-Length': 3}) + code=200, body=b'foo', headers={'Content-Length': 3}) self.mock_disks_and_gateways() locator = 'acbd18db4cc2f85cedef654fccc4a4d8+3+K@xyzzy' - self.assertEqual('foo', self.keepClient.get(locator)) + self.assertEqual(b'foo', self.keepClient.get(locator)) self.assertEqual('https://keep.xyzzy.arvadosapi.com/'+locator, - MockCurl.return_value.getopt(pycurl.URL)) + MockCurl.return_value.getopt(pycurl.URL).decode()) + @mock.patch('pycurl.Curl') + def test_head_with_remote_proxy_hint(self, MockCurl): + MockCurl.return_value = tutil.FakeCurl.make( + code=200, body=b'foo', headers={'Content-Length': 3}) + self.mock_disks_and_gateways() + locator = 'acbd18db4cc2f85cedef654fccc4a4d8+3+K@xyzzy' + self.assertEqual(True, self.keepClient.head(locator)) + self.assertEqual('https://keep.xyzzy.arvadosapi.com/'+locator, + MockCurl.return_value.getopt(pycurl.URL).decode()) class KeepClientRetryTestMixin(object): + disk_cache = False + # Testing with a local Keep store won't exercise the retry behavior. # Instead, our strategy is: # * Create a client with one proxy specified (pointed at a black @@ -821,7 +1205,7 @@ class KeepClientRetryTestMixin(object): # out appropriate methods in the client. PROXY_ADDR = 'http://[%s]:65535/' % (tutil.TEST_HOST,) - TEST_DATA = 'testdata' + TEST_DATA = b'testdata' TEST_LOCATOR = 'ef654c40ab4f1747fc699915d4f70902+8' def setUp(self): @@ -830,6 +1214,7 @@ class KeepClientRetryTestMixin(object): def new_client(self, **caller_kwargs): kwargs = self.client_kwargs.copy() kwargs.update(caller_kwargs) + kwargs['block_cache'] = self.make_block_cache(self.disk_cache) return arvados.KeepClient(**kwargs) def run_method(self, *args, **kwargs): @@ -843,7 +1228,9 @@ class KeepClientRetryTestMixin(object): def check_exception(self, error_class=None, *args, **kwargs): if error_class is None: error_class = self.DEFAULT_EXCEPTION - self.assertRaises(error_class, self.run_method, *args, **kwargs) + with self.assertRaises(error_class) as err: + self.run_method(*args, **kwargs) + return err def test_immediate_success(self): with self.TEST_PATCHER(self.DEFAULT_EXPECT, 200): @@ -867,7 +1254,8 @@ class KeepClientRetryTestMixin(object): def test_error_after_retries_exhausted(self): with self.TEST_PATCHER(self.DEFAULT_EXPECT, 500, 500, 200): - self.check_exception(num_retries=1) + err = self.check_exception(num_retries=1) + self.assertRegex(str(err.exception), r'failed to .* after 2 attempts') def test_num_retries_instance_fallback(self): self.client_kwargs['num_retries'] = 3 @@ -876,12 +1264,16 @@ class KeepClientRetryTestMixin(object): @tutil.skip_sleep -class KeepClientRetryGetTestCase(KeepClientRetryTestMixin, unittest.TestCase): +@parameterized.parameterized_class([{"disk_cache": True}, {"disk_cache": False}]) +class KeepClientRetryGetTestCase(KeepClientRetryTestMixin, unittest.TestCase, DiskCacheBase): DEFAULT_EXPECT = KeepClientRetryTestMixin.TEST_DATA DEFAULT_EXCEPTION = arvados.errors.KeepReadError HINTED_LOCATOR = KeepClientRetryTestMixin.TEST_LOCATOR + '+K@xyzzy' TEST_PATCHER = staticmethod(tutil.mock_keep_responses) + def tearDown(self): + DiskCacheBase.tearDown(self) + def run_method(self, locator=KeepClientRetryTestMixin.TEST_LOCATOR, *args, **kwargs): return self.new_client().get(locator, *args, **kwargs) @@ -919,13 +1311,58 @@ class KeepClientRetryGetTestCase(KeepClientRetryTestMixin, unittest.TestCase): (self.DEFAULT_EXPECT, 200)): self.check_success(locator=self.HINTED_LOCATOR) +@tutil.skip_sleep +@parameterized.parameterized_class([{"disk_cache": True}, {"disk_cache": False}]) +class KeepClientRetryHeadTestCase(KeepClientRetryTestMixin, unittest.TestCase, DiskCacheBase): + DEFAULT_EXPECT = True + DEFAULT_EXCEPTION = arvados.errors.KeepReadError + HINTED_LOCATOR = KeepClientRetryTestMixin.TEST_LOCATOR + '+K@xyzzy' + TEST_PATCHER = staticmethod(tutil.mock_keep_responses) + + def tearDown(self): + DiskCacheBase.tearDown(self) + + def run_method(self, locator=KeepClientRetryTestMixin.TEST_LOCATOR, + *args, **kwargs): + return self.new_client().head(locator, *args, **kwargs) + + def test_specific_exception_when_not_found(self): + with tutil.mock_keep_responses(self.DEFAULT_EXPECT, 404, 200): + self.check_exception(arvados.errors.NotFoundError, num_retries=3) + + def test_general_exception_with_mixed_errors(self): + # head should raise a NotFoundError if no server returns the block, + # and a high threshold of servers report that it's not found. + # This test rigs up 50/50 disagreement between two servers, and + # checks that it does not become a NotFoundError. + client = self.new_client() + with tutil.mock_keep_responses(self.DEFAULT_EXPECT, 404, 500): + with self.assertRaises(arvados.errors.KeepReadError) as exc_check: + client.head(self.HINTED_LOCATOR) + self.assertNotIsInstance( + exc_check.exception, arvados.errors.NotFoundError, + "mixed errors raised NotFoundError") + + def test_hint_server_can_succeed_without_retries(self): + with tutil.mock_keep_responses(self.DEFAULT_EXPECT, 404, 200, 500): + self.check_success(locator=self.HINTED_LOCATOR) + + def test_try_next_server_after_timeout(self): + with tutil.mock_keep_responses( + (socket.timeout("timed out"), 200), + (self.DEFAULT_EXPECT, 200)): + self.check_success(locator=self.HINTED_LOCATOR) @tutil.skip_sleep -class KeepClientRetryPutTestCase(KeepClientRetryTestMixin, unittest.TestCase): +@parameterized.parameterized_class([{"disk_cache": True}, {"disk_cache": False}]) +class KeepClientRetryPutTestCase(KeepClientRetryTestMixin, unittest.TestCase, DiskCacheBase): DEFAULT_EXPECT = KeepClientRetryTestMixin.TEST_LOCATOR DEFAULT_EXCEPTION = arvados.errors.KeepWriteError TEST_PATCHER = staticmethod(tutil.mock_keep_responses) + def tearDown(self): + DiskCacheBase.tearDown(self) + def run_method(self, data=KeepClientRetryTestMixin.TEST_DATA, copies=1, *args, **kwargs): return self.new_client().put(data, copies, *args, **kwargs) @@ -935,28 +1372,107 @@ class KeepClientRetryPutTestCase(KeepClientRetryTestMixin, unittest.TestCase): self.check_exception(copies=2, num_retries=3) +class AvoidOverreplication(unittest.TestCase, tutil.ApiClientMock): + + class FakeKeepService(object): + def __init__(self, delay, will_succeed=False, will_raise=None, replicas=1): + self.delay = delay + self.will_succeed = will_succeed + self.will_raise = will_raise + self._result = {} + self._result['headers'] = {} + self._result['headers']['x-keep-replicas-stored'] = str(replicas) + self._result['headers']['x-keep-storage-classes-confirmed'] = 'default={}'.format(replicas) + self._result['body'] = 'foobar' + + def put(self, data_hash, data, timeout, headers): + time.sleep(self.delay) + if self.will_raise is not None: + raise self.will_raise + return self.will_succeed + + def last_result(self): + if self.will_succeed: + return self._result + else: + return {"status_code": 500, "body": "didn't succeed"} + + def finished(self): + return False + + def setUp(self): + self.copies = 3 + self.pool = arvados.KeepClient.KeepWriterThreadPool( + data = 'foo', + data_hash = 'acbd18db4cc2f85cedef654fccc4a4d8+3', + max_service_replicas = self.copies, + copies = self.copies + ) + + def test_only_write_enough_on_success(self): + for i in range(10): + ks = self.FakeKeepService(delay=i/10.0, will_succeed=True) + self.pool.add_task(ks, None) + self.pool.join() + self.assertEqual(self.pool.done(), (self.copies, [])) + + def test_only_write_enough_on_partial_success(self): + for i in range(5): + ks = self.FakeKeepService(delay=i/10.0, will_succeed=False) + self.pool.add_task(ks, None) + ks = self.FakeKeepService(delay=i/10.0, will_succeed=True) + self.pool.add_task(ks, None) + self.pool.join() + self.assertEqual(self.pool.done(), (self.copies, [])) + + def test_only_write_enough_when_some_crash(self): + for i in range(5): + ks = self.FakeKeepService(delay=i/10.0, will_raise=Exception()) + self.pool.add_task(ks, None) + ks = self.FakeKeepService(delay=i/10.0, will_succeed=True) + self.pool.add_task(ks, None) + self.pool.join() + self.assertEqual(self.pool.done(), (self.copies, [])) + + def test_fail_when_too_many_crash(self): + for i in range(self.copies+1): + ks = self.FakeKeepService(delay=i/10.0, will_raise=Exception()) + self.pool.add_task(ks, None) + for i in range(self.copies-1): + ks = self.FakeKeepService(delay=i/10.0, will_succeed=True) + self.pool.add_task(ks, None) + self.pool.join() + self.assertEqual(self.pool.done(), (self.copies-1, [])) + + @tutil.skip_sleep -class RetryNeedsMultipleServices(unittest.TestCase, tutil.ApiClientMock): +@parameterized.parameterized_class([{"disk_cache": True}, {"disk_cache": False}]) +class RetryNeedsMultipleServices(unittest.TestCase, tutil.ApiClientMock, DiskCacheBase): + block_cache = False + # Test put()s that need two distinct servers to succeed, possibly # requiring multiple passes through the retry loop. def setUp(self): self.api_client = self.mock_keep_services(count=2) - self.keep_client = arvados.KeepClient(api_client=self.api_client) + self.keep_client = arvados.KeepClient(api_client=self.api_client, block_cache=self.make_block_cache(self.disk_cache)) + + def tearDown(self): + DiskCacheBase.tearDown(self) def test_success_after_exception(self): with tutil.mock_keep_responses( 'acbd18db4cc2f85cedef654fccc4a4d8+3', Exception('mock err'), 200, 200) as req_mock: self.keep_client.put('foo', num_retries=1, copies=2) - self.assertTrue(3, req_mock.call_count) + self.assertEqual(3, req_mock.call_count) def test_success_after_retryable_error(self): with tutil.mock_keep_responses( 'acbd18db4cc2f85cedef654fccc4a4d8+3', 500, 200, 200) as req_mock: self.keep_client.put('foo', num_retries=1, copies=2) - self.assertTrue(3, req_mock.call_count) + self.assertEqual(3, req_mock.call_count) def test_fail_after_final_error(self): # First retry loop gets a 200 (can't achieve replication by @@ -967,4 +1483,39 @@ class RetryNeedsMultipleServices(unittest.TestCase, tutil.ApiClientMock): 200, 400, 200) as req_mock: with self.assertRaises(arvados.errors.KeepWriteError): self.keep_client.put('foo', num_retries=1, copies=2) - self.assertTrue(2, req_mock.call_count) + self.assertEqual(2, req_mock.call_count) + +@parameterized.parameterized_class([{"disk_cache": True}, {"disk_cache": False}]) +class KeepClientAPIErrorTest(unittest.TestCase, DiskCacheBase): + disk_cache = False + + def tearDown(self): + DiskCacheBase.tearDown(self) + + def test_api_fail(self): + class ApiMock(object): + def __getattr__(self, r): + if r == "api_token": + return "abc" + elif r == "insecure": + return False + elif r == "config": + return lambda: {} + else: + raise arvados.errors.KeepReadError() + keep_client = arvados.KeepClient(api_client=ApiMock(), + proxy='', local_store='', + block_cache=self.make_block_cache(self.disk_cache)) + + # The bug this is testing for is that if an API (not + # keepstore) exception is thrown as part of a get(), the next + # attempt to get that same block will result in a deadlock. + # This is why there are two get()s in a row. Unfortunately, + # the failure mode for this test is that the test suite + # deadlocks, there isn't a good way to avoid that without + # adding a special case that has no use except for this test. + + with self.assertRaises(arvados.errors.KeepReadError): + keep_client.get("acbd18db4cc2f85cedef654fccc4a4d8+3") + with self.assertRaises(arvados.errors.KeepReadError): + keep_client.get("acbd18db4cc2f85cedef654fccc4a4d8+3")